i use paramAnalysis and how to get the output?

Zfeng,

Use value(v("vout") 'lin 10u) if you want to retrieve the curve for a value of a specific design variable. Normally if you don't specify the variable name (just the value to measure at), it takes that measurement at the inner sweep. When doing a parametric analysis of a transient, the inner sweep is the time axis, and the next outer sweep is the parametric variable. So you could get the values you want by doing:

value(value(v("vout") 50u) 10u)
value(value(v("vout") 50u) 20u)

Regards, 

Andrew.

Hi Andrew,

I posted a thread elsewhere regarding the picking of data during a parametric sweep using Skill code and calculating data.

I think I've found the reason behind why I'm getting only one set of results when a parameter is swept.  The parameter in question is "case"...as in process corner...this is in our PDK.

 But isn't 'case' a reserved construct/command?

 If so, I won't be able to use say a cross() function to pull parametric data where 'case' is one of the nested parameters.  Is it possible to alias the parameter in the skill code?

 Thanks in advance

As far as I know, there's nothing special about the word  "case". If you call sweepNames() it should show you the names in the resulting data when the simulation has finished...

Andrew.

Thanks Andrew

 Indeed, the sweepNames() function returns "case" "tsh" "time".

I can even autoplot the individual cases/corners.  I just can't seem to wrok on the data.  Here's a snapshot ..{fout is a declared and opened file handle with the 'w' option}

paramAnalysis("case" ?values '(0 1)

paramAnalysis("tsh" ?values '(-5e-10 -4e-10)

))

resultsDir("./psf1")

paramRun()

openResults("./psf1")

selectResults('tran)

tf=cross(v("Vout") 1 3 'falling)

ocnPrint(?output fout tf)

close(fp)

 ...........................END

Fingers crossed I've copied all the brackets across carefully.  If I ocnPrint v("Vout"), then I it prints the Vout vs time for both case 0 and 1.  If I change the desVar( "case" 0) to desVar( "Case" 0), then the above code prints both cases, but such a variable doesn't exist so the corner never moves from nominal.

I think I may go back to basics and try a nested 'foreach' loop.  I read that paramAnalysis() is quicker to implement and easier, but I can't figure out why it has a problem with 'case'.

Cheers Man!

So what's the actual problem? What you describe above sounds as if it's working correctly...

Andrew.

But it's not working correctly.  It only prints the 'tf' to the filename for case=0 only.  It can't work on the individual case data....or I should say, "case" parameter swept data other than case=0  When I plot tf vs tsh, I should see 2 distinct plots where the 'case' variable has changed.  instead, the plot shows one coincident on the other and looking at the data, I only see data in the printed file for one case. 

Hmm, that seems different from what you said earlier, when you said that ocnPrint was showing it for both case values. Maybe I misunderstood you.

Can you log a service request with customer support? I think we'd need to see this. What version are you using (I don't think you said)?

Andrew.

Spectre isr 10.1.1.056.isr5

ICFB 5.10.41.500.6.147

NOTE:  If you read that post earlier which confused you, you’ll see the word ‘case’ has  a capital C.  No such variable exists in my testbench, but it does print correctly to the the file. Just doesn’t changes anything because the variable ‘Case’ doesn’t exist.  ‘case’ however does (with a lower case c).

Lotta ‘cases’ flying around!

I take it I need to log this formally?  I’ll probably need to go through our CAD group for this.

It was the statement "If I ocnPrint v("Vout"), then I it prints the Vout vs time for both case 0 and 1." which made me think that it was working, because this seemed no different from the situation where you used desVar("Case"...). 

Yes, please log this as a service request via support.cadence.com - I think we'll need to take a look at the actual data.

Regards,

Andrew.